Diagnostic testing intensity for Legionnaires' disease: Spatio-temporal assessment and its effect on surveillance...

The global, national and subnational geographical and temporal distribution of reported Legionnaires' disease is extremely heterogeneous, and it is unknown to what extent this accurately reflects the variation in true disease incidence. We studied how diagnostic testing intensity varied across Danish municipalities between 2014 and 2022, how it influenced epidemiological surveillance, and how testing-adjusted reporting can be used to improve the study of disease incidence and its determinants.

We used data from the clinical Danish Microbiology Database and the Epidemiological Surveillance System and considered a tentative causal model of how Legionella infections give rise to the observed Legionnaires' disease surveillance data. We fitted spatio-temporal models using an approximate Bayesian inference for latent Gaussian models (INLA), providing probabilistic estimates. These allowed us to identify areas of increased risk and spatio-temporal interaction. Our assessment of the Legionnaires' disease testing intensity in Denmark showed considerable spatio-temporal variation across the country. The estimated municipal annual testing intensity ranged from 128 to 2,446 persons receiving at least one Legionella urinary antigen or PCR test per 100,000 inhabitants. The median increased between 2014 and 2022 steadily from 275 to 620 tested persons per 100,000 inhabitants, reflecting an upward trend. The proportion of tested persons with age over 70 increased from 44.8% to 56.4%. Increasing testing intensity leads to higher case reporting, until testing intensity reaches approximately 1000-1200 tested persons per 100 000 inhabitants. The estimated municipal annual testing-adjusted case reporting ranged from 1.4 to 12.0 per 100 000 inhabitants. The median fluctuated over the study period between 2.5 (range 1.4-6.2) in 2014 and 5.2 (range 1.5-11.7) in 2022 with a flat overall time effect.

We obtained estimates of the spatio-temporal variation of Legionnaires' disease among Danish municipalities. We quantified the positive effect of testing intensity on Legionnaires' disease reporting and found a threshold of annually testing slightly over 1% of the population above which the yield of new cases does not further increase. Despite limitations and possible bias, our study of testing-adjusted case reporting suggests that no substantial increase in Legionnaires' disease has occurred over the nine-year study period. Instead, case ascertainment by physicians has improved considerably through increased Legionella testing, particularly in elderly patients. Insight in the variation of testing intensity and its effect on Legionnaires' disease reporting can be used to improve guidance for Legionnaires' disease diagnosis, to better study determinants of Legionnaires' disease, and ultimately to improve Legionnaires' disease prevention and control.